It has been the goal of this work to establish a process chain for the characterisation of hyperspectral sensors beginning with the definition of the measurement procedures, followed by the automatic execution of the measurements and the analysis of the recorded data. A substantial goal was the automation of the characterisation measurement process. For this an operational concept was developed. It groups the necessary steps into three scopes of functions: the control of the CHB devices (Slave); the execution of the measurements (Master); and the control of the sensor (Sensor). The subdivision of the modules and the definition of software interfaces for the modules enabled a parallel and independent development of the three modules. Furthermore the individual modules can be modified independently from each other. This was particularly for the CHB control module an advantage (Slave) during testing the software interface of APEX at an early stage. In addition the “emulation” mode of the Slave module was very helpful for the interface tests. The APEX sensor has been characterised several times in the CHB facility. Thus one of the essential aims of the laboratory automation is fulfilled.
Regarding the aim of automating the CHB, this has been shown to be possible and has subsequently been implemented (all devices can now be controlled automatically). The necessary manual interactions are limited to the switching on and off of the devices. An extension of the module with further devices is unproblematic. This was demonstrated with the implementation of an additional filter wheel of the monochromator. The currently implemented Sensor modules are for AISA, ROSIS, ARES and the lock-in amplifier in combination with a photo diode (used for calibration of the monochromator). These examples show the usability of the sensor part of the operational concept.
The central part of the automatic execution of measurements is the Master module. It offers many additional options besides the execution of the measurements. For example the data transfer functions, the execution of programs or system commands both directly or in the background. In particular the possibility of executing programmes (e.g. raw data conversion) in the background function (parallel job) offers additional time saving potentialities.
The second main goal was the definition of generic measurement and analysis methods for spectral, geometric and radiometric characterisation for hyperspectral sensors. For each of the three measurement groups the measurement setups were described. The required devices and their parameters were defined. The defined sensor independent measurement methods have been successfully applied for the both test sensors. The various control commands for the different measuring tasks, consisting of settings for the CHB devices and the sensor, could be collected, and consequently it is now possible to perform the measuring sequences automatically. Measuring tasks over many hours have been executed without any problem and with only a few interactions of the CHB operator at the beginning of the measuring tasks.
7 Conclusions
The developed analysis procedures are generic after the conversion of the raw data into the generic data format (ENVI/IDL) since the raw data formats are sensor specific. The analysis procedures can be performed fully automatically due to their parameterisation. The Master records all sensor and laboratory device settings (e.g. sensor name, viewing angle) of the measurements which are necessary for the analysis procedures. Erroneous measurements caused by saturation, or incomplete measurements, can be easily identified on the basis of the analysis logs. The generated plots are useful for the recognition of artefacts. For each sensor three files were created which contain the analysis results of all spectral, geometric and radiometric measurements. The performed characterisation measurements and analyses for ROSIS and AISA showed the applicability of the sensor independent measurement and analysis methods.
Main accomplishments:
• A complete process chain has been established for the characterisation of hyperspectral sensors.
• The full automation of the CHB facility was implemented and the time effort to carry out the measurements decreases up to 50% depending on the measurement type. The manual interactions of the operator are reduced to a minimum (switching on and off of the devices, alignment of the sensor).
o The Master and Slave software modules are expandable by the addition of new sensors and devices with minimal effort.
o The Master module offers a high flexibility for the execution of most diverse characterisation measurements.
• Standard measurement and analysis methods for regular spectral, geometrical and radiometric characterisation for imaging spectrometers in the wavelength range from 0.4 µm up to 2.5 µm have been defined and were verified with two sensors (AISA and ROSIS).
Outlook
The planed upgrade of the ROSIS system (e.g. controlling system of the internal mirror, extension of the disk space, new graphic user interface) represents an improvement for future characterisation measurements. The following aspects should be considered for the design of future sensors: sufficient disk storage, a suitable graphical user interface and functions and features for a proper alignment of the sensor (e.g. alignment mirror) and possibilities for a direct data processing.
An important point for the future is an error estimation of the laboratory. The accuracies of the single devices in this thesis have been provided by the manufacturers. These have to be improved and further investigations on error propagation are planed for the future.
Acknowledgements
This work has been carried out with the support of scientist of DLR and from several other individuals which work at different institutions.
My special thanks go to Prof. Klaus I. Itten, supervising this work and Dr. Peter Gege for continuous support. Many thanks go also to Dr. Peter Haschberger, Jochen Fries, and Dr. Horst Schwarzer for additional feedback on this work. Special thanks go to Dr. Georg Wagner and Dr. Daniel Hoja for reading parts of the manuscript and motivation talks. I am grateful to Willem Vreeling supporting me with laboratory measurements. Sincere thanks also to the people from RSL for sharing knowledge and fruitful discussions, especially Francesco Dell’Endice. I thank all other people at DLR/IFM and temporarily present students for their contributions and the nice working ambience at the IFM.
Very special thanks are due to my husband and my mother for their encouragements and the facilitation to create this work.
Appendices
Appendix A. Technical information
Optical Bench
Figure A.1 show the opening of the DLR universal adapter where a sensor can be mounted on. The flanges have hole pattern as Leica’s PAV30 [106].
a) Maximum opening of 443 mm b) Adapter flanges Figure A.1: DLR universal adapter with rings
Figure A.2 show technical drawings of the off-axis parabolic mirror at the exit slit of the monochromator.
a) Reflectance of bright rhodium
b) Reflection of an off-axis paraboloidal
c) Off-axis parabolic mirror dimensions
Appendix A
Geometric measurement components
The layout of the vertical and horizontal slits on the slit wheel is illustrated in Figure A.3. The
slits 1 to 3 are the vertical slits and slit 4 to 5 are the horizontal ones.
Figure A.3: Layout of the vertical and horizontal slits on the wheel [108] Radiometric measurement components
The possible lamp combinations of the large integrating sphere are listed In Table A.1 and Figure A.4 shows the arrangement of the lamps in the large integrating sphere. Two power units operate two lamps (see lamp 8 and lamp18 as well as lamp 2 and lamp 10).
Table A.1: Lamp combinations of the large sphere Combination
no.
Complete
wattage Rack no. Lamp no. Lamp combination
1 90 2 2, 10 2*45 W 2 135 6, 2 2,10,6 3*45 W 3 180 2, 18 2,10,8,18 4*45 W 4 200 3, 11 3,11 2*100 W 5 300 3,11,17 3,11,17 3*100 W 6 400 3,7,11,17 3,7,11,17 4*100 W 7 600 1,9,14 1, 9, 14 3*200 W 8 800 1,9,11,14,17 1, 9,11,14,17 3*200 W + 2*100 W 9 1000 1,5, 9 ,11,14,17 1, 5,9,11,14,17 4*200 W + 2*100 W 10 1200 1,3 ,5 ,7,9,11,14,17 1, 3, 5, 7, 9, 11, 14, 17 4*200 W + 4*100 11 1625 1,2 ,3 ,5 ,6 ,7,9,11,12, 14,17,18 1, 2, 3, 5, 6, 7, 8, 9, 10, 11, 12, 14, 17, 18 5*200 W + 4*100 W + 5*45 W 12 2025 1,2,3,4,5,6,7,9,11,12, 13,14,17,18 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 17, 18 7*200 W + 4*100 W + 5*45 W Home position 5 1 6 2 3 4
Appendix A
Figure A.4: Lamp arrangement of the large sphere
Small integrating sphere
Figure A.5 shows the radiance and the uncertainty of the small integrating sphere. The sphere was calibrated at PTB in November 2007. The uncertainty (U) is 1% in the wavelength range from 390 nm to 1700 nm. The uncertainty increases in the shorter and longer wavelength range (see top of Figure A.5).
Appendix A 400 800 1200 1600 2000 2400 Wavelength (nm) 00 100 100 200 200 300 300 400 400 500 500 600 600 700 700 R ad ia nc e (m W m -2 n m -1 s r -1 ) 00 55 10 15 U ( % )
Figure A.5: Spectral radiance and uncertainty of the small sphere, calibrated in Nov. 2007 [34] Table A.2: Absolute diffraction efficiencies of the gratings
Wavelength (nm) Absolute diffraction efficiencies (%)
430 40 - 45 500 70 - 75 600 70 - 75 700 50 800 40 - 45 900 35 - 40
Appendix B. Additional Software Information
Table B.3: INI-files of the modules Slave, Master and SensorModule Content
Slave Network settings, hardware COM ports
Environmental data thresholds, default values for the emulation mode Setup times for the different devices
Monochromator correction equations Lamp combination list of the large sphere Speed of the axes of the folding mirror
Master File names and name spaces of the different XML schemata
File names of the different log files
Network settings for the communication with the Slave module IDL settings
Default parameter values for the different sensors
Sensor(s) COM port, timeout and directory
Table B.4: Procedures used in the spectral characterisation analysis process
Name of the IDL procedure Description
chbmo00_015_get_logs_data Reads the measurement data and Slave and Sensor log data
chbmo00_008_read_slaveorsensordata: file:monochromator_settings.txt
Reads the monochromator wavelength calibration values
chbmo00_017_read_powermeter_measurements Reads the monochromator output values chbmo00_016_calculate_pixel_number Calculate the spatial pixel number
according to the viewing angle
chbm03_004_gaussfit Gaussian function
chbm00_014_print_wmf Print Gaussian curves of 5 channels
Table B.5: Procedures used in the geometric characterisation analysis process (across and along track LSF)
Name of the IDL procedure Description
chbmo00_018_read_textfile input file: chbmo01_001_log.txt
Read analysis log template
chbmo00_015_get_logs_data Reads the measurement data and Slave and Sensor log data
chbmo00_008_read_slaveorsensordata: file:slitwheel_settings.txt
Reads the collimator and slit wheel settings
chbmo03_002_get_environment_data Read the laboratory ambient data chbmo00_017_read_powermeter_measurements Reads the monochromator output values chbmo00_016_calculate_pixel_number Calculate the spatial pixel number
according to the viewing angle
chbm03_004_gaussfit Gaussian function
Appendix B
Table B.6: Procedures used in the radiometric characterisation analysis process
Name of the IDL procedure Description
chbmo00_015_get_logs_data Reads the measurement data and Slave and Sensor log data
chbmo00_008_read_slaveorsensordata Reads the transmittance values of the neutral density filter
chbmo00_017_read_powermeter_measurements Reads the monochromator output values chbmo00_024_radiometric_dark_corrected_cube Corrects the nadir measurement data by
subtracting the dark current data and create a cube file
chbm03_004_gaussfit Gaussian function
chbm00_014_print_wmf Print Gaussian curves of 5 channels
Figure B.6 illustrates an example of the Slave software interface. The figure shows all names and data types of the relevant parameters of the “sensor” device group.
Figure B.6: Parameters and data types of the Slave interface group "sensor" ([55])
IDL functions
The IDL Gaussian function “GAUSSFIT” was modified by a return value for the status of the fit function. This value is used for error control. The used IDL function, implemented in IDL, is: (B-1)
σ
λ
A1 z= − whereλ
: wavelength (nm) 1A
: centre of the Gaussian function (nm)
σ
: standard deviation of the Gaussian function (nm)3 2 0
)
2
exp(
*
)
(
A
z
A
f
λ
=
−
+
Appendix B
(B-2) where
λ
: wavelength (nm)
A
0: height of the Gaussian function (DN)A
3: constant term (DN)Master module
The names, units and descriptions of the interface parameters of the Slave are listed in Table B.7.
Table B.7: Slave interface parameter list Device
group Parameter Units Description
pressure hPa returns the air pressure
temperature °C returns the temperature inside the laboratory humidity % returns the humidity inside the laboratory weathersta_updatetime ms measurement time cycle of the weather station ambient_light lux returns the ambient light
amlight_updatetime ms measurement time cycle of the ambient light software_mode - auto=automatic, demo=demonstration light - room light status: 0=no light, 1 = light can be on comments - comments from the operator
date - returns the date
time - returns the time (local)
measurement_time sec planned measurement time
setup_time sec estimated time for the setup of the devices request_id - id of the current Master request
status - returns the status of whole CHB hardware: 0=off, 1=on
error_code - error code
global
error_message - error message
sensor_name - specify sensor (AISA, ROSIS, APEX, …) cal_type - calibration spec=spectral, mode: geo=geometrical, rad=radiometric,
pol=polarisation
scan_angle ° scan angle of the instrument scanangle_ accuracy ° returns the scan angle accuracy
x_axis_offset ° x-axis difference between the home position of the mirror and the optical axis of the sensor
y_axis_offset ° y-difference between mirror home position and sensor optical axis pitch_angle_offset ° pitch angle of the sensor relative to nadir roll_angle_offset ° roll angle of the sensor relative to nadir
sensor
Appendix B
Device
group Parameter Units Description
height mm distance between folding mirror and sensor slit (altitude of entrance pupil) cal_mode - calibration geo=geometrical mode: spec=spectral, filterarm_angle ° mounted pol=polarisation filter filter: none=no filter, filter_number number of polarisation filter
filter_orientation ° orientation of the polarisation filter
mirror_yposition mm y-position of the mirror relative to home position mirror_angle ° angle of the mirror relative to nadir position status - returns the status of the folding mirror: 0=off, 1=on
error_code - error code
folding mirror
error_message - error message
turret - Returns the number of the turret
lamp_source - lamp type
lamp_voltage V settings power-supply of the lamp lamp_current A settings power-supply of the lamp shutter - shutter status: 0=close, 1=open
wavelength nm wavelength
filter_number - position of the filter in the wheel
entrance_slit_type - entrance mot=motorised slit type: man=manual, entrance_slit_width µm entrance slit width in 2 µm steps
entrance_slit_height mm entrance slit height 15 mm
exit_slit_type - exit slit type: man=manual, mot=motorised exit_slit_width µm exit slit width
exit_slit_height mm exit slit height, manual slit 1, 2, 4, 10 exit_filter - position of the filter in the wheel
grating - number of the grating
lines 1/mm returns the lines per mm blaze nm returns the blaze wavelength
status - returns the status of the monochromator: 0=off, 1=on
error_code - error code
Monochro- mator
error_message - error message slit_number - slit number
slit_angle ° angle of current slit relative to wheel home position
lamp_source - lamp type
lamp_voltage V settings power-supply of the lamp lamp_current A settings power-supply of the lamp
input - input device (other can be e.g. photo diode) status - returns the status of the collimator: 0=off, 1=on
error_code - error code
collimator
error_message - error message
diameter m diameter of the integrating sphere lamp_combination - See Table A.1
filter_list - list of the used transmission filters lamp_list - list of the used lamps
integrating sphere
Appendix B
Device
group Parameter Units Description
diode_updatetime ms measurement time cycle for photodiode status - returns the status of the integrating sphere: 0=off, 1=on
error_code - error code
error_message - error message
Figure B.7 shows the GUI for sensor settings for the ROSIS sensor in the Master software. In general the IP address of the computer is necessary for the Master software, on which the sensor commodity offers its service. Each implemented sensor has its own dialog because every sensor has other settings. For example, AISA has a shutter that can be closed; ROSIS has no shutter but a specific mirror position for dark measurements. The Master software opens the necessary GUI of the selected sensor.
Appendix B
Figure B.8: AISA sensor configuration GUI
The dialog for the step action collection is shown in Figure B.9. It is possible to collect, to delete or to change actions. The order of the step actions can be changed with the “Move up” or “Move down” button on the right side by marking the chosen action. The flow sequence of the actions is thereby from top to bottom.
Appendix B
Figure B.10: Master logging dialog
Figure B.11 illustrates the GUI for the selection and execution of measurements.
Figure B.11: Master GUI for the execution of measurements
The different data transfer possibilities of the sensor data in the Master module are listed in Table B.8.
Table B.8: Data transfer possibilities in the Master Software:
Local Copy
The copy command can be used if the Sensor and Master software runs on the same computer or if the data folder is shared on the sensor computer. Parameters of the copy command are the source and the destination path.
File Transfer Protocol (FTP)
If the Sensor and Master software are running on different PCs the recorded data can be copied with FTP. The Master uses the “GNU Wget [109]” program for the access and copy of data. Needed parameters are the user name and its password on the FTP server, the IP address and port number as well as the source path and destination path.
Hypertext Transfer Protocol (HTTP)
The Master software uses the existing “GNU Wget [109]” program for the data transfer. The needed parameters are the IP address of the sensor computer, the HTTP port number and the source and destination path.
Appendix B
Slave module
The content of the Slave observer function is shown in Figure B.12. The web page is divided into three parts. The left side comprises the requested device settings and the right side contains the current settings of the devices as well as error information. The frame at the bottom shows the current messages for the operator.
Figure B.12: Content of the web page of the Slave Observer
Figure B.13 illustrates the GUI for the monitoring of the environmental data. The monitoring can be switched on as well as the alert function for the illuminance meter and weather station.
Appendix B
Appendix C. Analysis Results
Table C.9 contains the wavelength calibration values of the monochromator. The values were determined for the gratings of turret 2 during the monochromator calibration in March 2008 [110].
Table C.9: Monochromator calibration values for the gratings of turret 2 (calibrated 2008-03-26) Grating no. Offset (nm) Gain Standard deviation (nm)
1 -0.08105 3.3227 · 10-4 0.142
2 -0.23935 1.9384 · 10-4 0.105
3 0.35757 -1.2918 · 10-4 0.077
Geometric Characterisation Results
0 5 10 15 20 25 30 35 40 45 50 55 60 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 Channel number F W H M s ta n d a rd d e v ia ti o n ( m ra d * 1 0 -2 )
a) pixel 1 b) pixel 1 a) pixel 2 b) pixel 2 d) pixel 364 e) pixel 364 Figure C.14: Standard deviation of the FWHM values of the different measurements and pixels
Appendix C
AISA across track LSF measurement results
Table C.10 contains the across track measurement results for AISA at nine spatial positions (see chapter 6.4.2).
Table C.10: AISA across track LSF measurement results of channel 250 and different spatial pixels Pixel number Viewing angle (°) FWHM (°) FWHM (mrad) Sampling distance (°)
1 25.745 0.192 3.351 2 25.668 0.258 4.503 0.077 3 25.559 0.303 5.288 0.109 72 16.527 0.276 4.817 73 16.391 0.273 4.765 0.136 113 10.930 0.278 4.852 114 10.793 0.273 4.765 0.137 152 5.520 0.277 4.835 153 5.379 0.274 4.782 0.141 191 0.068 0.289 5.044 192 -0.072 0.283 4.939 0.140 230 -5.388 0.271 4.730 231 -5.526 0.269 4.695 0.138 269 -10.777 0.259 4.520 270 -10.913 0.258 4.503 0.136 310 -16.357 0.281 4.904 311 -16.493 0.277 4.835 0.136 362 -23.154 0.284 4.957 363 -23.281 0.282 4.922 0.127 364 -23.407 0.280 4.887 0.126
The following figure contains the AISA FWHM standard deviations for four spatial pixel and all channels consistent to the viewing angles plotted in Figure 6.22.
Appendix C 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 Channel number S ta n d a rd d e v ia ti o n F W H M ( 1 0 -2 m ra d )
Pixel 73 Pixel 113 Pixel 153 Pixel 192
Figure C.15: AISA standard deviation for FWHM of the across track LSF for several pixels
The following picture is an exemplary measurement result of pixel 153 and channel 196.
Appendix C
Spectral Characterisation Results
Figure C.17 illustrates the analysis results of measurement number 1 of Table 6.7 of AISA. The difference between the centre wavelengths of the intensity corrected (419.766 nm) and non intensity corrected (419.773 nm) is 0.007 nm. The FWHM values are 1.528 nm (uncorrected) and 1.579 nm for the intensity corrected.
100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 418 418.5 419 419.5 420 420.5 421 421.5 422 Wavelength (nm) S ig n a l (D N )
Raw data Intensity corrected raw Gaussian fit raw data Gaussian fit intensity corrected
Figure C.17: Comparison of channel 1 none corrected and intensity corrected data of AISA (pixel 192)
0 100 200 300 400 500 600 700 800 900 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 Wavelength (nm) S ig n a l (D N )
raw c 120 raw c 121 raw c 122 raw c 123 raw c 124